Abstract: In view of the shortage of harvesting and transportation labor force in citrus orchards and the difficulty of realizing autonomous navigation, this paper designs a follow navigation method to realize the automatic transportation of citrus. Based on the fusion of lidar and visual information, this paper uses the HSV threshold to segment the image to obtain the direction of the picking guide robot. The marker point cloud of the guide robot is clustered and identified in the direction range to obtain the relative pose between the transportation following robot and the picking guide robot. The transportation robot is controlled through an algorithm to follow and complete autonomous navigation. The system is tested in a simulated environment. The results show that at a maximum speed of 0.5 m/s, the average longitudinal deviation is 1.5 cm, the average transverse deviation is 1.0 cm, and the average heading angle deviation is 1.107° during straightline following. The system continues to follow the target under an arc track, and the parking test of the following robot between lines also meets the working requirements. The system provides technical support for autonomous transportation machinery in citrus orchards and lays a foundation for practical work in the orchard.

Key words: citrus, vision, lidar, data fusion, following navigation

摘要： 针对柑橘园采收运输劳动力缺乏以及自主导航方式实现难度较大的问题，设计跟随导航的方式实现柑橘的自动化运输。基于激光雷达和视觉信息融合，采用HSV阈值分割图像，获取采摘引导机器人的方向，并在方向范围内聚类识别引导机器人标志物点云，获得运输跟随机器人与采摘引导机器人之间的相对位姿，并通过控制算法使运输机器人进行跟随完成自主导航。在模拟环境中对该系统进行测试，结果系统在最大0.5m/s的速度下，直线跟随时平均纵向偏差1.5cm，平均横向偏差1.0cm，平均航向角偏差1.107°，弧线轨迹下持续跟随目标，跟随机器人在行间的停车试验也基本满足工作要求。系统为柑橘自主运输机械提供了技术支撑，为实际果园工作奠定基础。

关键词: 柑橘, 视觉, 激光雷达, 数据融合, 跟随导航